Project Lambda, a project made for Western AI 2022/2023 and presented at CUCAI 2023.

Project Lambda consists of two AI agents built to play the first-person shooter video game ”Counter-Strike: Global Offensive” (CS:GO). The first agent uses a combination of YOLOv7 object detection and A* pathfinding algorithm. The second agent uses a deep neural network trained on a large data set of video footage of professional CS:GO matches using behavioural cloning and offline reinforcement learning. While most ”aim-bot” programs tap into the games memory to retrieve information such as the coordinates of players and aim the mouse directly at them, the goal for our project was to create an AI that can match the performance and behaviour of real players while relying purely on the visuals on the screen. We look to compare the behaviour and realism of an AI built with pathfinding vs an AI built with machine learning.

Paper: [ Ссылка ]

GitHub: [ Ссылка ]

Music by [ Ссылка ]

свернуть